MQA Deep Dive - I published music on tidal to test MQA

[Zensō]

Unfortunately, you are misreading what I wrote. I have never written that "all 16-bit recordings have only 13 bits of real data," nor would I.

In the example I gave of a real-world recording, the noise level was sufficiently high in the very low bass that the music above that noise would not require all 16 bits. However, as I also wrote, the noise level decreased with increasing frequency, and at higher frequencies 16 bits would not be sufficient to capture the music signal.

John Atkinson
Technical Editor, Stereophile

John, can you describe in non-technical terms what you see as the benefits of MQA to the average music listener?

Thanks..

 

[mansr]

John, can you describe in non-technical terms what you see as the benefits of MQA to the average music listener?

I'd settle for _any_ meaningful terms.

 

[Atanasi]

In the example I gave of a real-world recording, the noise level was sufficiently high in the very low bass that the music above that noise would not require all 16 bits.

Does MQA support customized shaping according to the noise spectrum of the recording? The test tones demonstrated here produced most noise at high frequencies, the opposite to your example. Actually, noise at low frequencies is less useful to encode information, because bandwidth is narrower compared to high frequencies.

 

[mansr]

Does MQA support customized shaping according to the noise spectrum of the recording? The test tones demonstrated here produced most noise at high frequencies, the opposite to your example. Actually, noise at low frequencies is less useful to encode information, because bandwidth is narrower compared to high frequencies.

The MQA dither noise is fixed.

 

[John Atkinson]

The test tones demonstrated here produced most noise at high frequencies, the opposite to your example.

Of course. The test tone spectra are showing the effect of noiseshaping. My example showed the spectrum of environmental and electrical noise in a 24-bit recording of "room tone". The two should not be confused.

John Atkinson
Technical Editor, Stereophile

 

[sandymc]

Unfortunately, you are misreading what I wrote. I have never written that "all 16-bit recordings have only 13 bits of real data," nor would I.

In the example I gave of a real-world recording, the noise level was sufficiently high in the very low bass that the music above that noise would not require all 16 bits. However, as I also wrote, the noise level decreased with increasing frequency, and at higher frequencies 16 bits would not be sufficient to capture the music signal.

John Atkinson
Technical Editor, Stereophile

MQA is using those three bits for its hidden channel all the time, regardless of what is playing.

I would agree that you could create a perceptual encoding scheme that in effect used less bit depth for low frequency-only sections of audio, and upped resolution for high frequency sections.Which I assume is what MQA is trying to do. Without knowing the details of the MQA encoding scheme, it appears to me that 16-bit MQA is an attempt to use the top 13 bits to keep red book compatibility, while using the bottom 3 bits to store variable length perceptual encoding information. All very clever, but at it's heart that is an attempt to combine a fixed frequency encoding scheme (the top 13 bits) with a variable frequency encoding scheme (the bottom 3 bits). You would expect that to lead to all sorts of compromises, which is what the tests suggest is the case.

 

[John Atkinson]

John, can you describe in non-technical terms what you see as the benefits of MQA to the average music listener?

Thanks..

I have already referred ASR readers to the articles I have written on MQA. They are easily accessible at Stereophile's website. I don't see why I am obliged to repeat what I have written here.

John Atkinson
Technical Editor, Stereophile

 

[Raindog123]

The MQA dither noise is fixed.

But it’s not even about dither. It’s about their “triangle” or “tilted noise floor” claim. While that concept - of using some of lower frequency bits (where those bits are “not needed due to lower dynamic range“) - looks cool when you draw that “origami unfold sketch” in log scale... But in reality, they claim that a few (3? variable?) bits of those lower frequencies (<1kHz) can represent the entire unfolded band (of 20kHz)... Definitely a bandwidth issue.

 

[abdo123]

My guess: As the noise floor in those test files was not dithered and couldn't be read as noise at some steps of the process, the algorithm enters a situation that it is not programmed for. As I said in the previous post: garbage in -> garbage out.

The person who submitted the files admitted that some of his files were indeed dithered and the problem remained nonetheless. I guess it serves the point you're making to purposefully ignore that fact.

 

[mansr]

But it’s not even about dither.

It is though, because that is what gives undecoded MQA the characteristic spectrum.

 

[levimax]

while using the bottom 3 bits to store variable length perceptual encoding information.

As well as other undisclosed DRM like / Watermark like information.

 

[TurbulentCalm]

We will need one of these:
View attachment 125547

I just started playing the soundtrack to Dune by Toto and then scrolled the page to see this.

Coincidence can be such and amazing thing.

Then I remembered what was left behind on the floor, after the Navigator was push away: could this be a good analogy for MQA’s three bits of dithered ultrasonic origami, hidden below the floor of the Navigator’s habitat?

 

[Raindog123]

It is though, because that is what gives undecoded MQA the characteristic spectrum.

I am not arguing against your point. I (and @Atanasi) are bringing an additional one.

 

[muslhead]

Well, since you’re here arguing for MQA, I thought you might use the opportunity to clarify your current position for those of us who are less technically inclined.

His "current" position was stated 3 years ago and as far as i can search on Stereophile, not been updated even though his position was based upon some recently proved MQA false statements.

 

[Zensō]

I have already referred ASR readers to the articles I have written on MQA. They are easily accessible at Stereophile's website. I don't see why I am obliged to repeat what I have written here.

John Atkinson
Technical Editor, Stereophile

Well, since you’re here arguing for MQA, I thought you might use the opportunity to clarify your current position for those of us who are less technically inclined.

 

[mansr]

Well, since you’re here arguing for MQA, I thought you might use the opportunity to clarify your current position for those of us who are less technically inclined.

He's here doing damage control for MQA while also trying to drive some traffic to his site. Do not expect a meaningful discussion.

 

[Zensō]

He's here doing damage control for MQA while also trying to drive some traffic to his site. Do not expect a meaningful discussion.

Yeah, I know, I just thought I’d see if he was up for getting real for a moment. Crazy me.

 

[TurbulentCalm]

Here is my HDTracks folder alone:

View attachment 125561

218 albums I have bought with vast majority of them high-res. No DRM. No control. No nothing.

Wow! Just think hot much room you could save if you compressed all those audio files with MQA.

 

[Raindog123]

Yeah, I know, I just thought I’d see if he was up for getting real for a moment. Crazy me

IMHO, we still need them. To argue, no matter how stupid, selective, and past-each-other those discussions are. As this is one of our few weapons remain: to continue asking questions and exposing nonsense - creating smoke signals and hoping that those will get noticed by [streaming] investors and execs. What’s the alternative? I doubt it will be an exciting discussion amongst one-camp chaps.

 

[RichB]

Unfortunately, you are misreading what I wrote. I have never written that "all 16-bit recordings have only 13 bits of real data," nor would I.

In the example I gave of a real-world recording, the noise level was sufficiently high in the very low bass that the music above that noise would not require all 16 bits. However, as I also wrote, the noise level decreased with increasing frequency, and at higher frequencies 16 bits would not be sufficient to capture the music signal.

John Atkinson
Technical Editor, Stereophile

How many bit are required for this sample recording?
A MQA CD or equivalent download would degrade your sample recording.
Nothing less, not one bit less, than CD quality 44.1/16 bits is lossless CD qualtiy.

Tidal Homepage:

The base package at $9.99 is 320K MP3.

The next package is HiFi:

Lossless Quality 1411 Kbps
TIDAL HiFi allows you to stream audio using the lossless format FLAC, creating a crisp and robust music streaming experience.

What is HiFi quality and bit-rate is not CD quality.

Their top-tier is Master
High-Res Quality 2304-9216 Kbps
Master sound quality is authenticated and unbroken, with the highest possible resolution — as flawless as it sounded in the mastering suite and precisely as the artist intended.

Tidal does not promise CD quality nor Hi-Res quality, instead using a nebulous term "Master".
The entire site seemed to avoid claiming any real quality metric.
Obviously, MQA tracks are not "precisely as the artist intended" but the claim is legally valid, because artists like Neil Young have left Tidal. So clearly, those left, must agree.

They are clearly not offering the "highest possible resolution" but it is their highest possible resolution.
I love this " as flawless as it sounded in the mastering suite and precisely as the artist intended" beauty.
There are many terrible masters and they the above statement is valid, but really.
Tidal and MQA are a match made in heaven, full of spin and deception.

QOBUZ Homepage.

QOBUZ lowest plan is for CD quality audio at 44.1 kHz.
There could be MQA CD lossy but only if the label delivers it.
Here are the QOBUZ quality provided

• Hi-Res audio up 24 bit up to 192kHz
• CD quality 44.1/16 bits
• MP3 320 kbps

QOBUZ only uses kbps for MP3 when it is a clear quality metric.

I think you must agree with these statements:

1. Tidal MQA delivered in 44.1 kHz FLAC files are not CD quality in the dynamic range may be lost.
2. Tidal MQA does not claim to deliver Hi-Res audio.
3. Master quality audio is nebulous and has no meaning whereas Hi-Res audio does.

Buried deep in your posts and articles there is usefully information, deep and skillfully worded.

- Rich